[gdal-dev] Add Mercator_variant_A method?

[Even Rouault]

Le lundi 06 avril 2015 23:09:09, Brad Hards a écrit :
> On Mon, 6 Apr 2015 01:56:16 PM you wrote:
> > Le lundi 06 avril 2015 13:26:11, Brad Hards a écrit :
> > > The problematic part is that "PROJECTION" definition. I was originally
> > > going to argue it should be Mercator_1SP, but it looks like the EPSG
> > > decided to change it (in EPSG::2010.058) to this new
> > > "Mercator_variant_A" form, and keep the "Mercator_1SP" form as an
> > > alias.
> > 
> > > Should we add the new form, with a change like:
> > The issue with this is that you only fix the exportToProj4(). But there
> > are other places in the code base where SRS_PT_MERCATOR_1SP is tested
> > (for example when creating a dataset in other formats). It would be
> > annoying to test in all those places both variants. Why not
> > "normalizing"
> > "Mercator_variant_A" into "Mercator_1SP" on ingestion of GeoPackage WKT ?
> > This would be a bit similar to all the tricks we do when ingesting ESRI
> > WKT.
> 
> Do you mean just a test in gpkg/ogrgeopackagedatasource.cpp, or a more
> general normalisation since presumably that will come up in other WKT?

Brad,

I'm not sure if we have many other sources that could return 
Mercator_variant_A (is there some OGC standard that normalizes this name?), so 
for now in GPKG driver would probably be OK. And if later we discover that 
other drivers can return it, we could create a 
OGRSpatialReference::morphFromModernEPSG() (or any better name!) method for 
that purpose.

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com